> Here if locked == true but say !(sb->s_flags & SB_ACTIVE), we fail to > unlock the superblock now AFAICT. Yeah, I've already fixed that up in-tree. I realized this because I've fixed it correctly in the last patch. > And here if you really mean it with some kind of clean bail out, we should > somehow get rid of the s_active reference we have. But exactly because of > that getting super_lock_excl() failure here would be really weird... > > Otherwise the patch looks good. With the above fix folded in can I take your Ack?
